Supplementary data for <a 
href=http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=15479714 > "Bayesian analysis of signaling networks governing embryonic stem cell
fate decisions" </a> by Woolf PJ, Prudhomme W, Daheron L, Daley GQ, 
Lauffenburger DA. <br><br>
<a href=BayesNet.tar>overview of the software with examples and documentation in the README.pdf file (updated 2/25/2005)</a> <br>
<a href=1.and.2.parent.plots.nb.gz>1.and.2.parent.plots.nb.gz </a> <br>
<a href=Supl.Table.1.pdf>Supl.Table.1.pdf </a> <br>
<a href=Mouse.Data.suppl.xls>Mouse.Data.suppl.xls </a> <br>
<a href=RUNNING.tar.gz>source code </a><br>
<br>
Data analysis with the source code provided on this page:<br><br>

1) Provide raw data in a format similar to Janes.Pre.xml (in the bnj.2.5.gaussfilt folder).  Here you specify the names of your variables, their structural priors, and the raw data.  Discretization takes place in the next step.
<br><br>
2) Discretize data using the program gaussfilt.  For example, from the folder bnj.2.5.gaussfilt, the run the unix command:
<br><b>nohup java -Xmx200M gaussfilt/gaussfilt Janes.Pre.xml Janes.test.5.100.xml 100 5 > o &</b>
<br><br>
This command runs java with 200 MB of memory.  Runs the program gaussfilt (Gaussian filter) on the "pre-dataset" (e.g. non sampled set) Janes.Pre.xml.
This dumps a file called Janes.test.5.100.xml using 100 samples for 
each entry binned into 5 bins.
<br><br>
3) Score networks using the program K2 (as a side note, this is no longer the K2 algorithm, but used to be when in the BNJ package.  Now it is a monte-carlo greedy hill climber with mutiple restarts)
For example, from the folder bnj.9.29.2003, I can run the command:
<br><b>java -Xmx500M k2/K2 Janes.10.200.xml test.xml</b>
<br><br>
That learns from the input Janes.10.200.xml and dumps an output to  
test.xml <br><br>
To run this program in the background for a long time, a nohup is helpful.  For example,<br>
<b>nohup java -Xmx500M k2/K2 Janes.10.200.xml test.xml > output.txt & </b><br><br>
prevents the program from quitting when you log off, and dumps output to the file output.txt.  The "&" command runs the program in the background.